Design And Implementation Of Tensile Strength Prediction System For Medium And Heavy Plate Products

With the continuous development of intelligent manufacturing and the integration of informatization and industrialization in China's iron and steel industry,industrial software manufacturing execution system(MES)and production marketing system(ERP)are the core elements to support manufacturing enterprises achieving intelligent manufacturing.A large amount of operation activities related data for years is accumulated and stored in system.The valuable information data can be mined and extracted by mining technology.Important factors affecting product quality and cost can be found out by analyzing data,so as to optimize process,reduce defects,reduce costs and improve product yield,thus lay the foundation for intelligent production.Product performance of Ansteel medium and thick plate is predicted and analyzed in this paper.All factors affecting product performance such as product performance parameters,production process and chemical composition are focused research.Through collecting and integrating data related to production operation of all levels information systems,building plate theme data pool,normativing data standard,setting up plate production process of 360 panoramic data view and realizing data resource sharing.Advanced big data analysis technology is used to construct data multidimensional analysis,association analysis,prediction analysis and other models,extracting and in-depth mining at multilevel data,so as to realize the transformation from data to information and then to knowledge.SPSS Modeler platform is used to input data sample of data pool into 4 data mining models : classification and regression tree(CART),linear regression,generalized linear and artificial neural network model,through repeated training and validation prediction model and combined with product process actual situation for overall evaluation,finally,we can draw the conclusion that artificial neural network model prediction is more accurate and more suit practical production process.This prediction model can directly reflect the factors affecting of medium and thick plate properties.In this paper,through establishing medium and thick plate product performance prediction platform,selecting the proved artificial neural network analysis model with most optimal prediction accuracy.Operator input chemical composition value,key process parameters index and heating temperature value to numerical simulate the whole process of plate production,and system will figure out plate performance target value according to the input parameters,to predict plate performance.It has important theoretical and guiding significance to improve product quality,reduce cost and R&D new products.